The following used ffmpeg to detect and decode files, and sox to adjust gain.
The following shell script will detect HDCD files that were ripped bit-perfect (EAC/XLD).
#!/bin/sh -
ffmpeg -hide_banner -nostats -y -v verbose -i "$1" -t 30 -af hdcd -f s24le a.out 2>&1
One can pass options to the HDCD decoder to create an audio file that has a change in tone when an HDCD feature is enabled on the track. I then run that output file through sox to create a spectrogram to quickly inspect to see where the feature was active. For analyze_mode, use either lle (max_adjust_gain) or pe (peak_extend). Additional information can be found in the ffmpeg documentation.
ffmpeg -i input.m4a -af hdcd='analyze_mode=lle' -acodec pcm_s24le out.wav
sox out.wav -n spectrogram
Once you have found an album that should be decoded, the following will loop over a directory (CD rip) of files and convert, decoding HDCD on each file. It is only really worth doing this on files that have peak_extend: enabled permanently. I've seen some files with max_gain_adj set, but have yet to see it actually show up in the audio in any meaningful manner.
This command works in the fish shell. I ripped to ALAC with the .m4a extension. If you ripped to FLAC, you should use a different extension.
for i in *.m4a
set name (echo $i | cut -d'.' -f1)
ffmpeg -i $i -af hdcd -acodec pcm_s24le $name-24.wav
end
First, one needs to find the peak value on the album:
sox *.wav -n stats
This will process through all of the converted files in the directory and find the peak level of the whole group. You could test each file individually if you wanted...
To adjust the gain, find the peak value from the previous command. I personally knock it down by 0.1 dB. If the peak was -4.6 dB, I would apply 4.5 dB of gain in this step.
for i in *-24.wav
set name (echo $i | cut -d'.' -f1)
sox $i $name-out.wav gain 4.5
end
Now, just convert your files back to the desired encoding and retag them.